Cognitive Self Training
by @zrxparley
Daily cognitive training, dream review, dream-scene narration, and self-improvement loop for OpenClaw, Hermes, Codex, Claude Code, and other AI agents. Use w...
Prefer a project-local store so learning remains scoped and inspectable:
bash cognitive-self-training/scripts/init_cognitive_training.sh .
If the skill is installed elsewhere, run the script from that installed skill path. If scripts are unavailable, create the structure in references/storage-schema.md. For dream scheduling details, read references/dream-protocol.md and references/automation.md. For dream description style configuration, read references/dream-styles.md.
Use this root selection:
1. If an OpenClaw workspace exists for the current task, use ~/.openclaw/workspace/.cognitive-training/.
2. Else use the current project/workspace root: .cognitive-training/.
3. Ask before writing to a global home directory, sharing across sessions, or modifying AGENTS.md, SOUL.md, TOOLS.md, MEMORY.md, CLAUDE.md, or .github/copilot-instructions.md.
Never record secrets, tokens, private keys, raw environment dumps, full private transcripts, health data, or third-party personal data. Store short redacted summaries instead.
clawhub install cognitive-self-training